Skillv1.0.4

ClawScan security

Skill Guangjiao · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousApr 30, 2026, 9:22 AM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill's declared runtime instructions (which require a TRADEGPT_API_KEY and write/read local skill data) are coherent with its Canton Fair lead‑mining purpose, but the published registry metadata (which lists no required env vars or config paths) is inconsistent with the embedded SKILL.md and clawhub.yaml — this mismatch and a few minor scope questions warrant caution before installing.
Guidance: This skill is an instruction-only Canton Fair lead‑mining helper that calls yunlvai's API. Before installing: - Confirm the TRADEGPT_API_KEY requirement: the skill files declare this key but the registry summary omitted it — ask the publisher to clarify why metadata differs. - Only provide the API key if you trust https://api.yunlvai.com and understand what data is sent to it; rotate the key if you later revoke access. - If you enable the optional Messaging API, review what credentials that will require and where messages are routed. - Review the declared local storage path (./data/yunlv-skills/guangjiaoFAI/) and ensure it is acceptable for storing contact data; verify retention/cleanup in your environment. - Be mindful of privacy/regulatory obligations (GDPR/CAN-SPAM) before importing or contacting harvested leads. - Because the package is documentation-only (no install), risk is limited to API calls and local file writes — if you need higher assurance, request the publisher to correct registry metadata and provide an explicit privacy/data-processing statement or sample network logs showing only expected API endpoints are contacted.

Review Dimensions

Purpose & Capability: noteThe skill claims to query a '云旅AI MatchGPT API' to retrieve exhibitor data and to generate outreach content; requiring an API key (TRADEGPT_API_KEY) and optionally a user-configured messaging API is consistent with that purpose. However, the top-level registry summary provided earlier stated 'Required env vars: none' and 'Required config paths: none', while both SKILL.md and clawhub.yaml declare TRADEGPT_API_KEY and list API endpoints — an internal inconsistency in the package metadata.
Instruction Scope: okSKILL.md is instruction-only and clearly limits its actions: it reads its own reference files, calls the yunlvai API for data, performs local processing and writes results to a declared ./data/yunlv-skills/guangjiaoFAI/ directory. Instructions do not ask to read unrelated system files or exfiltrate secrets. It does reference an optional 'Notification/Messaging API' that a user may configure to deliver messages, which is reasonable but should be configured deliberately.
Install Mechanism: okThere is no install spec and no code files beyond documentation — lowest install risk. Nothing will be downloaded or extracted by the skill itself according to provided files.
Credentials: noteThe skill requires a single primary credential (TRADEGPT_API_KEY) to call its backend API, which is proportionate to its described functionality. The inconsistency is that registry-level metadata reported no required env vars while the skill files declare the API key; that discrepancy should be resolved. The optional Messaging API (user-configured) could require additional credentials if enabled — the user should be aware of that when configuring it.
Persistence & Privilege: okThe skill does write data to a skill-specific local path (./data/yunlv-skills/guangjiaoFAI/) as documented in SKILL.md; it does not request always:true, system-level privileges, or modifications to other skills. Persisted leads and logs are within the skill's stated scope and are described with a 7-day cleanup policy, though the platform cannot verify enforcement.